The Effects of Coaching Subjects in the Recognition of Facial Expressions

A wide range of individual differences in the ability to identify facial expressions of emotion has been reported. As evidence that these inequalities are due in large part to differences of practice in reacting to expressive criteria, rather than to innate factors, F. H. Allport has offered the data from an experimental observation of the effects of training subjects of varying ability for fifteen minutes in the analysis of facial expressions. The gain in score made by his subjects on the facial expression test after the training period was roughly inversely proportional to their original standing; i.e., the less proficient persons gained the most and the more proficient gained the least. Repeating this procedure with larger groups of subjects, both with the same and longer study periods, the writer largely confirmed Allport's data. However, a control experiment, affording no opportunity for study, yielded similar data, except that on the average no gain in score resulted. The inverse relationship between original ability and amount of gain or loss should not, therefore, be attributed to the effects of training. Further analysis shows that when the reliability of the facial expression test, found by retesting a group of subjects, is taken into account, the negative coefficient practically disappears. The negative coefficient, upon which Allport bases his theory, is thus found to result in this study from the erroneous assumption that the reliability of the facial expression test is perfect. The writer, though inclined to agree with Allport's theory, concludes that this type of experimental evidence, when closely scrutinized, fails to support the theory.